The effect of awareness at the medium access control layer of vehicular ad-hoc networks

(1)

by

Marthinus Johannes Booysen

Dissertation presented for the degree of Doctor of Philosophy

in the Faculty of Engineering at Stellenbosch University

Supervisors:

Prof. Gert-Jan van Rooyen

Department of Electrical & Electronic Engineering, Stellenbosch University.

Prof. Sherali Zeadallly

Department of Computer Science and Information Technology University of the District of Columbia.

(2)

Declaration

By submitting this dissertation electronically, I declare that the entirety of the work con-tained therein is my own, original work, that I am the sole author thereof (save to the extent explicitly otherwise stated), that reproduction and publication thereof by Stellen-bosch University will not infringe any third party rights and that I have not previously in its entirety or in part submitted it for obtaining any qualication.

December 2013

Date: . . . .

(3)

Abstract

The hidden terminal problem, coupled with high node mobility apparent in vehicular networks, present challenges to ecient communication between vehicles at the Medium Access Control (MAC) layer. Both of these challenges are fundamentally problems of lack of awareness, and manifest most prominently in the broadcasting of safety messages in infrastructure-free vehicle-to-vehicle communications.

The design of existing contention-free and contention-based MAC approaches gener-ally assumes that nodes that are in range of one another can take steps to coordinate communications at the MAC layer to overcome the hidden terminal problem and node mobility. Unicasting with the existing MAC standard, IEEE 802.11p, implicitly assumes an awareness range of twice the transmission range (a 1-hop awareness range) at most, since handshaking is used. For broadcasting, the assumption implies an awareness range that is at most equal to the transmission range, since only carrier sensing is used. Existing alternative contention-free approaches make the same assumption, with some protocols explicitly using a 1-hop awareness range to avoid packet collisions. This dissertation chal-lenges the convention of assuming that a 1-hop awareness range is sucient for networks with high mobility, such as VANETs.

In this dissertation, the impact of awareness range and management of the awareness information on MAC performance is researched. The impact of the number of slots that is required to support the awareness range is also evaluated.

Three contention-free MAC protocols are introduced to support the research. The rst is an improved version of an existing MAC method, which is used to demonstrate the eects on performance of changes to awareness management. The second MAC uses three competing processes to manage awareness information. The second MAC is designed for a congurable awareness range and congurable number of slots, and is used to evaluate the eects of awareness range and number of slots on MAC performance. The third MAC is random access based and is used to evaluate the impact on performance of removing awareness completely. An analytical model is developed to support the simulated results. The simulation results demonstrate that awareness range, awareness information man-agement, and number of slots used are key design parameters that signicantly impact on MAC performance. The results further show that optimal awareness-related design parameters exist for given scenarios.

Finally, the proposed contention-free and random access MAC methods are simulated and performance compared with IEEE 802.11p. All three outperform the contention-based standard IEEE 802.11p.

(4)

Uittreksel

Die versteekte-nodus-probleem, gekoppel met die hoë vlakke van nodusbeweging teen-woordig in voertuignetwerke, bied uitdagings vir doeltreende kommunikasie tussen vo-ertuie in die medium-toegangbeheer- (MAC) vlak. Beide van hierdie probleme spruit uit beperkte bewustheid, en manifesteer veral in die uitsaai van veiligheidsboodskappe in infrastruktuurvrye voertuig-na-voertuig-kommunikasie.

Die ontwerp van bestaande wedywerende en nie-wedywerende MAC benaderings neem aan dat nodusse wat binne bereik van mekaar is, stappe kan neem om kommunikasie op die MAC-vlak te koördineer, ten einde probleme met versteekte nodusse en mobiliteit te oorkom. Vir punt-tot-puntkommunikasie met IEEE 802.11p, impliseer dié aanname 'n be-wustheidstrekking van hoogstens twee keer die radiobereik (1-hop bebe-wustheidstrekking), aangesien bladskud gebruik word. In die geval van uitsaai, impliseer die aanname 'n be-wustheidstrekking hoogstens gelyk is aan die radiobereik, aangesien slegs draeropsporing gebruik word. Nie-wedywerende metodes maak dieselfde aanname, met sommiges wat eksplisiet 1-hop-bewustheidstrekking gebruik om pakkieverliese te voorkom. Hierdie ver-handeling wys dat hierdie aanname nie geld vir netwerke met hoë mobiliteit nie, soos wat die geval is vir VANET.

In hierdie verhandeling word die impak van bewustheidstrekking en bestuur van die bewustheidsinligting in die MAC-vlak ondersoek. Die impak van die aantal tydgleuwe wat nodig is om die bewustheidstrekking te ondersteun word ook ondersoek.

Drie nie-wedywerende metodes word bekendgestel om die navorsing te ondersteun. Die eerste is 'n verbeterde weergawe van 'n bestaande MAC, wat gebruik word om die eekte van bewustheidsbestuur op MAC-werkverrigting te beoordeel. Die tweede MAC is ontwerp om veranderbare bewustheidstrekking en hoeveelheid tydgleuwe te ondersteun, en word gebruik om die eekte van bewustheidstrekking en hoeveelheid tydgleuwe op MAC werkverrigting aan te beoordeel. Die derde MAC is ewetoeganklik (onbewus van omliggende nodusse) en word gebruik om die impak van die verwydering van bewustheid op werkverrigting te ondersoek. 'n Analitiese model is ontwikkel om die simulasieresultate te ondersteun.

Die simulasieresultate dui aan dat bewustheidstrekking, bestuur van bewustheidsin-ligting, en hoeveelheid tydsgleuwe sleutel-ontwerpsveranderlikes is wat 'n beduidende im-pak het op MAC werkverrigting. Die resultate wys verder dat optimale ontwerpsveran-derlikes, in terme van bewustheid, bestaan vir gegewe scenario's.

Laastens, word die nie-wedywerende en ewetoeganklike MAC-metodes wat gesimuleer word se werkverrigting vergelyk met IEEE 802.11p. Al drie MAC metodes vaar beter as die wedywerende standaard, IEEE 802.11p.

(5)

Publications

Work in this manuscript has been published as follows:

M.J. Booysen, S. Zeadally, and G.-J. van Rooyen, "Survey of media access control protocols for vehicular ad hoc networks", IET Communications, vol. 5, no. 11, pp. 16191631, July 2011.

M.J. Booysen, J.S. Gilmore, S. Zeadally, and G.-J. van Rooyen, "Machine-to-machine (M2M) Communications in Vehicular Networks", KSII Transactions on Internet and Information Systems, pp.529546, vol. 6, no. 2, Feb 2012.

M.J. Booysen, S. Zeadally, and G.-J. van Rooyen, "A Performance Comparison of Media Access Control Protocols for Vehicular Ad-hoc Networks (VANETs)", IET Networks, pp. 1019, vol. 1, no. 1, March 2012.

M.J. Booysen, S. Zeadally, and G.-J. van Rooyen, "Impact of Neighbor Awareness at the MAC Layer in a Vehicular Ad-hoc NETwork (VANET)", IEEE Wireless Vehicle Conference (WiVeC), 23 June 2013, Dresden, Germany.

M.J. Booysen and G.-J. van Rooyen, "Performance Evaluation of Neighbor-Awareness at the Media Access Control (MAC) Layer for Vehicular Ad-Hoc Networks (VANETs), IEEE Intelligent Vehicles Symposium (IV), 2326 June 2013, Gold Coast, Australia.

(6)

Acknowledgements

I would like to express my sincere gratitude to the following people and organisations: Prof Gert-Jan van Rooyen for

a lovely beef burger in Jozi and many lunches around Stellenbosch his continued guidance and support, especially help with

* interpretation of the results * formulation of the hypotheses

* construction of the analytical models * the structure of this thesis

* multiple reviews Prof Sherali Zeadally for

Krispy Kreme donuts in DC

help with choosing the general topic

teaching me how to write academic material through endless hours of review giving me exposure to the academic research community

Dr Christoph Sommer and Prof Falko Dressler for chicken schnitzel on a snowy day in Erlangen support with Veins and SUMO

help with identifying the specic topic

The examiners for their valuable feedback and meticulous reviews The various reviewers of the published work for their suggestions

MIH for their nancial assistance towards this research and free lunches

The Electrical & Electronic Engineering department at Stellenbosch University for end of the month cake Friday

exibility and support

(7)

List of Figures

1.1 Vehicular communications network architecture. . . 3

2.1 Overview of the IEEE VANET standards. . . 14

2.2 IEEE VANET standards. . . 15

2.3 VANET and conventional DSRC frequency allocations. . . 15

2.4 Message structure in IEEE 802.11p. . . 16

2.5 Message structure in self-organising TDMA. . . 18

2.6 VeSOMAC message structure for three nodes. . . 19

2.7 MCTRP message structure. . . 21

2.8 CBMMAC message structure. . . 23

2.9 Hybrid medium access approach used by DMMAC. . . 24

2.10 Adaptive broadcast frame message structure in DMMAC. . . 25

2.11 Flow diagram for DMMAC. . . 25

2.12 Saturated steady-state SATs for DMMAC. . . 27

2.13 DMMAC information removal. . . 27

3.1 Simulation model. . . 34

3.2 SUMO graphical user interface . . . 35

4.1 RF ranges and awareness ranges for a two-directional highway scenario. . . 41

4.2 Highway area in-range of node n. . . 43

4.3 Number of in-range neighbours for dierent levels of receiver sensitivity (PRX). 44 4.4 Number of in-range neighbours for dierent values of the path loss exponent. . 45

4.5 MAMBA awareness control. . . 47

4.6 MAMBA refresh through collective memory. . . 48

4.7 Hidden node problem for the vehicular scenario with nRF= 8. . . 53

4.8 Hidden node problem induced by node mobility. . . 56

5.1 Timing arrangement used to compare MACs . . . 60

5.2 MAMBA receiver throughput. . . 60

5.3 MAMBA packet delivery ratios. . . 62

5.4 MAMBA latencies. . . 62

5.5 MAMBA over time for the highway scenario with high density trac. . . 64

5.6 MAMBA over time for the highway scenario with low density trac. . . 64

5.7 MAMBA over time for the urban scenario with high density trac. . . 65

5.8 MAMBA over time for the urban scenario with low density trac. . . 65

5.9 Packet delivery ratio for an average neighbour count of 4. . . 67

5.10 Packet delivery ratio (in 3D) for an average neighbour count of 4. . . 67

5.11 Packet delivery ratio for an average neighbour count of 10. . . 68 ix

(11)

5.17 Dynamic eects for an average neighbour count of 10. . . 71

5.18 Dynamic eects for an average neighbour count of 23. . . 71

5.19 Throughput for an average neighbour count of 4. . . 73

5.23 Latency for an average neighbour count of 4. . . 76

5.24 Latency for an average neighbour count of 10 . . . 76

5.27 The eect of awareness range for an average neighbour count of 4. . . 78

5.31 Neighbour-agnostic packet delivery ratio. . . 81

5.32 Neighbour-agnostic receiver throughput. . . 82

5.33 Neighbour-agnostic end-to-end latency. . . 82

B.1 Generic multi-server single queue system. . . 97

B.2 Multi-server queueing system state machine. . . 97

B.3 Markovian queue representation. . . 100

B.4 Markov chain model latency results. . . 102

(12)

List of Tables

2.1 IEEE standards for VANETs. . . 13

2.2 DSRC Channel allocation by the FCC. . . 14

2.3 CBMMAC channel categorisation. . . 22

2.4 Summary of recently proposed MAC protocols. . . 29

3.1 VANET simulation tools. . . 33

3.2 Simulation parameters used in our simulation tests. . . 38

4.1 Number of in-range neighbours (nRF)for dierent values of α and PRX. . . 45

5.1 Percentage throughput improvement of MAMBA. . . 61

5.2 Number of times a node was unable to nd an available slot. . . 61

5.3 Percentage improvement of MAMBA over DMMAC and IEEE 802.11p. . . 63

5.4 Congurations for maximum packet delivery ratio. . . 72

5.5 Congurations for optimal receiver throughput. . . 72

5.6 Eect of awareness range on MAC performance. . . 80

5.7 Comparison between IEEE 802.11p, DMMAC, and NYAMA. . . 81

(13)

Nomenclature

Acronyms

2G 2nd Generation International Mobile Telecommunications 3G 3rd Generation International Mobile Telecommunications

ABF Adaptive Broadcast Frame

ACK ACKnowledge

AIFS Arbitration Inter-Frame Spacing

AIS Automatic Identication System

AP Access Point

ARIB Association of Radio Industries and Businesses ASTM American Society for Testing and Materials

BSS Basic Service Set

CA Collision Avoidance

CALM Communications Access for Land Mobiles CBMMAC Clustering-Based Multichannel MAC

CCH Control CHannel

CDMA Code Division Multiple Access

CEN European Committee for Standardisation

CFR Code of Federal Regulations

CH Cluster Head

CRC Cluster-Range Control

CRD Cluster-Range Data

CRP Contention-based Reservation Period

CSMA Carrier Sense Multiple Access

CTS Clear To Send

CW Contention Window

DCF Distributed Coordination Function

DIFS DCF Inter-Frame Spacing

DMMAC Dedicated Multi-channel MAC

DSRC Dedicated Short Range Communication

EDCA Enhanced Distributed Channel Access

EDGE Enhanced Data Rates for GSM Evolution

ETC Electronic Toll Collection

(14)

FCC Federal Communications Commission FDMA Frequency Division Multiple Access

GPRS General Packet Radio Service

GPS Global Positioning System

GUI Graphical User Interface

ICC Inter-Cluster Control

ICD Inter-Cluster Data

IDM Intelligent Driver Model

ISM Industrial Scientic and Medical

ISO International Organisation for Standardisation

ITS Intelligent Transportation System

IVC Inter-Vehicular Communication

LLC Logical Link Control

MAC Media Access Control

MAMBA Medium Access through Memory Bifurcation and Administration MCTRP Multi-Channel Token Ring Protocol

MiXiM MiXed sIMulator

MLME MAC Layer Management Entity

NAV Network Allocation Vector

NSAF Non-Safety Application Frame

NYAMA Neighbour Yielding and Aware Medium Access

OBU On Board Unit

PHY PHYsical layer

PLME Physical Layer Management Entity

OMNET++ Objective Modular Network Testbed

QoS Quality of Service

RF Radio Frequency

RFN Ring Founding Node

RSU Road Side Units

RTS Request To Send

SAT Slot Allocation Table

SCH Service CHannel

SDMA Space Division Multiple Access

SNR Signal to Noise Ratio

SoTDMA Self-organising TDMA

SUMO Simulation of Urban Mobility

TC Technical Committee

TDMA Time Division Multiple Access

TIGER Topologically Integrated Geographic Encoding and Referencing

TraCI Trac Control Interface

TTL Time To Live

TXOP Transmit OPportunities

V2I Vehicle to Infrastructure

V2V Vehicle to Vehicle

VANET Vehicular Ad-hoc NETworks

Veins Vehicles in network simulation VeSOMAC Vehicular Self-Organising MAC

WAVE Wireless Access for Vehicle Environments

(15)

Wi-Fi Wireless Communication Based on IEEE 802.11 WiMAX Worldwide Interoperability for Microwave Access

WME WAVE Management Entity

(16)

List of symbols used

a Signal attenuation (path loss) [dB]

A Surface area of road in range of a node [m2]

B MAC transmit buer size [packets]

d Distance (transmission distance) [m]

dRF RF range (distance) [m]

H Number of hops used

N Total number of nodes

Nave Average number of active nodes

ni Node number i

nRF Number of in-range neighbours

NRX Receiving node

Nint Intermediate node

NTX Transmitting node

l Number of lanes

L Eective lane length in coverage area A [m]

Lw Lane width [m]

P Probability matrix, with pi,j as elements

Pi Probability vector with pi as elements

pi,j Probability of ni and nj choosing unique slots

pi Probability of ni choosing a unique slot from all its

in-range neighbours

PRX Receiver sensitivity [dBm]

PTX Transmitter power [dBm]

ri Packets received by node i

S Slots per TDMA frame (slots per SAT cycle)

SNRmin SNR threshold [dB]

si,j Nodes in range of node i and j

ti Packets transmitted by node i

tinf Information travel time [s]

tlat End-to-end latency [s]

tphy Physical travel time [s]

ti Packets transmitted by node i

Tsim Simulation duration [s]

Tslot Slot duration [s]

v Vehicle speed [m/s]

w1 distance from middle lane to nearest edge of opposite

side of highway [m]

w2 distance from middle lane to furthest edge of opposite

side of highway [m]

α Path loss exponent

λ Wavelength [m]

ρ Packet delivery ratio [%]

σ Vehicle density [vehicles/km/lane]

(17)

Chapter 1 Introduction: Medium access control

for vehicular ad-hoc networks

In the eld of vehicular communications, vehicles act as mobile nodes in wireless networks. In an ad-hoc network, nodes communicate directly with each other, without an Access Point (AP). Vehicular Ad-hoc NETworks (VANETs) are dierent from other ad-hoc net-works because of the high node mobility, the variable node density, and the unpredictable and harsh communications environment.

Vehicular communications comprise of two main modes of communication, Vehicle to Infrastructure (V2I) and Vehicle to Vehicle (V2V). The former designates communication between the On Board Units (OBUs) on vehicles and an infrastructure, through Road Side Units (RSUs). The latter refers to communication between vehicles that connect through OBUs. OBUs are network nodes mounted to vehicles and therefore inherently mobile and wireless. RSUs are stationary network nodes and are usually mounted in an elevated position on existing transportation infrastructure, such as trac lights, street lights and road signs [1]. RSUs provide a wireless link to vehicles and a wireless or wired link to the infrastructure. In this work we focus primarily on V2V communication, but relevant aspects of V2I communication are also covered where necessary.

The major drive for the development of vehicular communications has been the pleth-ora of foreseen applications, as well as the emergence of wireless networking technologies [2, 3,4, 5,6]. Applications for VANETs can be divided into the following broad categor-ies: safety related; trac management and transportation eciency; user infotainment services; and Internet connectivity [7, 6, 8]. Safety related applications include lane change assistance, cooperative forward incident warning, intersection collision avoidance, and emergency or incident warning [2]. Trac management applications form part of

(18)

a greater Intelligent Transportation System (ITS), which aims to make transportation systems smarter, faster, safer and more convenient. ITS functions include toll collec-tion, intersection management, cooperative adaptive cruise control, and detour or delay warning. The market impetus for VANETs is expected to be driven by user infotain-ment services rather than the safety or trac applications [7]; government and industry investments in safety and trac management development are based on this premise [9]. Infotainment applications range from multimedia delivery to email services and augmen-ted reality services [10].

Due to the wide variety of expected VANET applications, VANETs need to support a broad range of requirements. Safety messages require fast and guaranteed access to the wireless medium and a short transmission delay, and safety messages are relatively short [9]. The infotainment services could require a heavier data load, with less severe timing requirements [7]. For safety applications, a high Quality of Service (QoS) needs to be ensured, while for the user infotainment services this may not be a stringent requirement. Furthermore, for viable augmented reality applications, such as location information over-lay, substantial processing and accurate location determination will be required, while for trac management the requirement for these could be less stringent.

One view of the key components in a vehicular network, and how they interact, are indicated in Figure 1.1 from [11]. The safety and trac management services are separated from the entertainment and Internet connectivity in Figure 1.1. However, work done by Amadeo et al. has shown that these services can coexist without the support of existing wireless technologies such as WiMAX, Wi-Fi and cellular communications [12].

1.1 Vehicular connectivity with existing wireless

technologies

VANET solutions that have been developed using existing technologies (e.g. cellular phones or Wi-Fi) are not suciently robust to cope with the mobility inherent to VANETs [13, 15]. Limited connections currently exist between vehicles and dierent types of in-frastructure, such as the Internet and toll collection facilities. Whenever an occupant of a vehicle carries a mobile phone with a data connection to a cellular service, the user can access the Internet from within the vehicle. The cellular connectivity is dedicated to In-ternet connectivity and is therefore inherently infrastructure-based and cannot guarantee fast association, required by VANET [13, 15, 16]. Newer vehicles are also equipped with built-in mobile phone connectivity that enables various Internet-based services to occu-pants [17, 18]. These connections are, however, limited to Internet connectivity through infrastructure. The applications envisaged for VANET require fast association and in-frastructure independent communications. Vehicles are also connected to inin-frastructure such as Electronic Toll Collection (ETC) systems in numerous countries (for example, AUTOPass in Norway, Via Verde in Portugal, I-PASS in Illinois US, Salik in Dubai) with

(19)

RSU OBU

Internet

Intra-c

ar

Inte

r-c

ar

V2V

Ext

ra-car

_V2I

MM

Networks

Servers and

Consumers

ITS network

WIMAX Wi-Fi Cellular WAVE MOST ring CAN bus WAVE Servers and databases

Operators and end users

Navigation Emergency

detectors Audio-visual _sources Displays

Safety and traffic management Entertainment and Internet connectivity OBU RSU V2I V2V Wi-Fi CAN ISP ITS MM MOST WIMAX WAVE

Controller Area Network Internet Service Provision

Intelligent Transportation Systems MOST Master

Media Oriented Systems Transport

Worldwide Interoperability for Microwave Access Wireless Access in Vehicular Environments

On Board Unit

Road Side Unit (WAVE) Vehicle to Infrastructure Vehicle to vehicle Wireless Fidelity

Figure 1.1: Vehicular communications network architecture. Vehicular networks will initially use a hybrid of existing wireless technologies which include cellular technologies, Wi-Fi (IEEE 802.11a/b/g/n), WAVE (IEEE 802.11p), and WiMAX (IEEE 802.16) [7][13]. Vehicles Vehicular communication networks form part of Intelligent Transportation Systems (ITS), which aims to make transportation systems smarter, safer, and more convenient [14]. Vehicular communication networks stand to gain from the hierarchical and autonomous paradigms evident in Machine-to Machine (M2M) communications [11].

ranges of ten to fteen meters. The low bandwidth and relatively short range provided by ETC restricts its ability to host the VANET applications mentioned in this work.

(20)

1.2 Medium access control in the vehicular

environment

In any networking environment, one of the key aspects of the communication protocol stack is the Medium Access Control (MAC) layer. The MAC layer determines which node is given access to the shared wireless medium at what time.

1.2.1 MAC challenges in the VANET context

Several design challenges still need to be addressed at the MAC level to achieve fast, reliable, and fair access [19, 20]. A unique characteristic of a VANET that distinguishes it from other ad-hoc networks is the high mobility of nodes (vehicles). The greatest challenges in making VANET a cost-eective technology that can easily be deployed and adopted, is the time it takes to establish connections between vehicles (nodes) and other vehicles, or between vehicles and RSUs, as well as the delays incurred during the man-agement of access to the underlying wireless medium. The challenge for communication between vehicles (V2V communications) is that infrastructure-based centralised coordin-ation (e.g. from an RSU) is not available to manage and coordinate access to the wireless medium [21]. The key challenges are to determine which vehicle, if any, assumes con-trol of coordinating access to the medium on an ad-hoc basis, and how the time slots (if used) and channels are shared fairly between the vehicles that are in range of each other [21]. The problem is most noticeable in the scenario where two cars travel in opposite directions at speed, attempt to establish a connection, and need to transfer information on a constantly changing medium, while managing the medium and accompanying data collisions [7,19,22]. In that time, the nodes need to coordinate medium access and share the required information with accommodating surrounding nodes, and ensure that mes-sage delivery meets all the application requirements. The bottleneck in this process is the Medium Access Control (MAC) layer [7, 19, 23]. The work in [24] has shown that this challenge of coordination is not resolved with the existing VANET MAC standard, IEEE 802.11p.

Since wireless transceivers cannot transmit and receive simultaneously on the same frequency, collision detection is not as straightforward as with wired communication. Kenney lists the following important challenges of a MAC layer for VANET [25]:

The hidden terminal problem where two nodes are outside of each other's range, but both attempt to communicate with a node that is within the range of both. This problem is likely in pure V2V environments where there is no centralised com-munication coordination. The result of the hidden terminal problem is message collisions.

The dynamic nature of VANETs. Given the changing conditions and vehicle po-sitions, the frequency response and propagation delays of a channel could vary

(21)

signicantly.

VANETs need to be scalable, to ensure network performance in both low and high vehicle densities.

The dierent applications expected for VANETs, result in disparate requirements on the system. A guarantee is required that the safety messages will be delivered such that the receiver has sucient time to respond. Multimedia streaming applications require that the QoS is sucient to allow clear and decipherable delivery to the end user.

The hidden terminal problem is also identied as especially problematic by Sjöberg et al. in [26]. According to [19], the MAC should be robust against frequent disconnects between nodes, which could occur due to the highly mobile and varying nature of VANETs. Since nodes in a VANET are vehicles, which are inherently mobile, the MAC should be optimised for frequent disconnects and hand-os with other OBUs and RSUs [19]. The MAC layer needs to ensure fairness among all stations sharing the wireless medium and oer predictable access to the shared medium [20].

1.2.2 Proposed MAC approaches to support VANET

Many MAC approaches have been proposed for the vehicular environment [24,27,28,20, 29, 30]. These MAC methods can be broadly classied as based, contention-free, or a hybrid of these two [9]. Contention-based approaches rely on carrier sensing, back-o and retry schemes, while contention-free approaches rely on time division multiple access and synchronisation schemes.

In a contention-based approach (such as IEEE 802.11p) the nodes contend with each other for an opportunity to use the medium. Contention-free approaches use Time Divi-sion Multiple Access (TDMA), in which time slices, called slots, are allocated to nodes to ensure that transmissions are collision free each node transmits in a designated slot. Contention-based approaches perform better under low network trac loads and contention-free approaches perform better under heavy network trac loads [9]. MAC methods for the V2V environment need to be decentralised and independent of infra-structure, which is, by design, the case for contention-based MAC methods. However, for TDMA-based contention-free methods, this requirement for decentralised and independ-ent coordination implies the ability for vehicles to self-organise to coordinate slot alloc-ation. The problem with contention-based approaches is that unbounded delays could occur, which is unacceptable for the delivery of time-critical safety messages. For this reason, MAC methods that have been proposed as alternatives to IEEE 802.11p standard for the delivery of safety messages in the vehicular environment, have been contention-free approaches.

This dissertation focuses on contention-free approaches. The repeating sequence of a xed number of slots is called a frame, and to coordinate communication, a node owns

(22)

a slot in a frame. A key challenge for self-organising TDMA MAC methods, is the co-ordination and communication of slot ownership in this frame. Two approaches have been proposed to coordinate communications in contention-free MAC approaches. The rst approach is to make groups or clusters of nodes [27, 29]. In this conguration, one of the nodes in a cluster acts as a group coordinator and is responsible for maintaining slot allocation, as well as communicating slot allocation to the other nodes in the group. Since vehicles are inherently mobile, any node can go out of range of other nodes and the distribution of the group continuously changes, which means substantial overheads are required to reselect the coordinator, and to maintain the group information. Group co-ordination is further complicated by the fact all the nodes in a group are not necessarily in range of one another, making coordination and hand-over of coordination responsibilities problematic.

The second approach to coordinate communications in contention-free MACs, distrib-utes the coordination function to the nodes. With this approach, each node maintains and communicates its own, independent, list of TDMA slot allocations. We refer to this list as the Slot Allocation Table (SAT). In order to coordinate the communications, each node uses its slot to transmit its SAT, in addition to transmitting the data it wishes to transmit. The information in the SAT is dierent for dierent approaches: in the ap-proach proposed in [27] the SAT uses a bit for each slot, where a 0 means unoccupied, and a 1 means occupied. Since only a bit is used, the communication overhead is low, but the slot ownership becomes anonymous and intelligent SAT maintenance becomes a challenge. In the work in [24] and [30] the SAT contains an identier for the owners of each slot, which means the SAT can be maintained as nodes come within range and go out of range. In order to self-organise the slots, these contention-free MACs take into account the slot allocations of surrounding nodes. When receiving slot allocation information, the receiving node will incorporate this information into its own SAT, using some form of information management scheme.

1.3 Dissertation statements and hypotheses

Dissertation statement 1:

The awareness range of a MAC is a key determinant of its performance in a highly mobile network such as a VANET.

Since the hidden terminal problem is fundamentally a problem of lack of awarness, this work evalautes the eects of increased neighbour awareness on MAC performance. This dissertation refers to the number of hops used to collect MAC layer information from surrounding nodes, as the awareness range (see Section 4.2 on page 41). Nodes coordinate transmission based on their awareness of other nodes, which is limited by the awareness

(23)

range. The RF communications range is dened as the 0-hop awareness range.

The generic IEEE 802.11 MAC is a contention-based approach, which was designed for, and has been used extensively and successfully in networks with nodes that are static or have low mobility. Contention-based MAC design assumes that nodes that are in range of one another can take steps, such as handshaking or carrier sensing, to coordinate communications at the MAC layer. This assumption implies a 1-hop awareness range (i.e. awareness range just beyond the immediate neighbours) for unicast, where Request To Send/Clear To Send (RTS/CTS) handshaking is used, and an awareness range that is equal to the communications range for broadcast (where carrier sensing is used). The same assumption, that in-range nodes can coordinate access to the shared medium, is made in the design of contention-free MAC methods, although some approaches explicitly introduce a 1-hop awareness range to attempt to overcome the hidden node problem.

These assumptions may be valid for networks with no or low mobility, but in networks with high mobility, where the time it takes for coordination information to be transferred is in the same region as the time it takes to travel the physical distance between nodes (such as VANET), the awareness range has to be extended to beyond the immediate in-range neighbours, and even beyond the rst-hop neighbours to overcome the hidden node problem, and collisions caused by node mobility. Too much awareness, i.e. awareness of neighbour that is irrelevant to local communications, could also aect performance ad-versely.

Hypothesis 1.1:

There exists an awareness range that results in an optimal ratio of packets received to packets transmitted, for a given TDMA frame length.

Hypothesis 1.2:

There exists an optimal awareness range in terms of receiver throughput, for a given TDMA frame length.

Hypothesis 1.3:

The optimal awareness range is dierent for dierent in-range neighbour counts. Hypothesis 1.4:

An increase in awareness range beyond a threshold, increases end-to-end latency. Hypothesis 1.5:

A self-organising contention-free MAC method with limited awareness management can be improved with simple improvements in awareness management.

Hypothesis 1.6:

(24)

IEEE 802.11p in terms of ratio of successful packet delivery, receiver throughput, and end-to-end latency.

Hypothesis 1.7:

Removing awareness from a MAC has an eect on packet delivery ratio and receiver throughput, similar to decreasing the number of hops used.

To realise the benets of an increased awareness in a contention-free MAC, the chosen number of transmission slots per TDMA cycle must support the specic awareness range.

This dissertation refers to the number of slots per TDMA cycle as the Slot Allocation Table (SAT) size (The SAT is used for gathering slot usage information from surrounding vehicles, and its size is equal to the number of slots used). An increase in the awareness range requires an increase in the number of slots used, to accommodate the additional nodes in the awareness range, and to avoid saturation of the SAT. A decrease in the awareness range requires a decrease in the number of slots used, to avoid channel under-utilisation, and to ensure a fast enough cycle to overcome node mobility.

Hypothesis 2.1:

The packet delivery ratio approaches 100% as the SAT size increases at the cost of increased latency and reduced throughput.

Hypothesis 2.2:

There exists an optimal SAT size in terms of throughput for a given awareness range. Hypothesis 2.3:

The optimal SAT size is dierent for dierent optimal awareness ranges. Hypothesis 2.4:

Latency increases as the SAT size increases. Hypothesis 2.5:

Collisions of communication packets caused by node mobility can be prevented with a multi-hop awareness range and large enough SAT.

The existing standard for V2V communications, IEEE 802.11p MAC, is out-performed by non-contention-based MAC protocols with multi-hop

(25)

awareness ranges.

Three MAC protocols are proposed in this dissertation. A contention-free approach with 1-hop awareness range, a contention-free approach with multi-hop awareness range (and matching SAT sizes), and a neighbour-agnostic approach with random slot selection. All three approaches outperform IEEE 802.11p for all metrics used.

Hypothesis 3.1:

A self-organising contention-free MAC method with one-hop awareness outperforms IEEE 802.11p.

Hypothesis 3.2:

A self-organising contention-free MAC method with multi-hop awareness (and large enough SAT size to support the increased nodes int he awareness range) outperforms IEEE 802.11p.

Hypothesis 3.3:

A self-organising random access MAC method with no awareness, but with discrete time slots, outperforms IEEE 802.11p.

1.4 Research objectives

Research objective 1:

To simulate vehicular communications with focus on the MAC layer, on a represent-ative simulation platform. To achieve this objective, a simulation environment will be identied to perform the task of both mobility and communications simulation. The simulation environment with various representative trac models and commu-nications models will be set up and congured.

To modify an existing contention-free MAC method to evaluate performance improve-ment due to a change in the neighbour awareness manageimprove-ment scheme.

To develop a novel contention-free MAC method that uses a congurable awareness range and SAT size. The awareness range and SAT size will be congurable to eval-uate the eect of these. The method will be developed with a focus on awareness to perform better than the existing IEEE 802.11p standard.

(26)

To develop a mathematical model of the eects of awareness ranges and SAT size on the performance of the proposed MAC method.

To validate the simulation test results against the developed mathematical models.

1.5 Scope of the work

This dissertation focuses on Vehicle-to-Vehicle (V2V) communications. The dissertation evaluates the performance of Medium Access Control (MAC) layers for the broadcasting of safety messages. The MAC layers evaluated are the existing contention-based standard (IEEE 802.11p), a contention-free method called DMMAC, and three proposed MAC methods with dierent awareness management strategies. The eects on performance of awareness range (number of hops) and awareness memory (SAT size) are investigated. Highway and urban scenarios are considered, using high density and low density road trac. Various radio and signal propagation properties are used to conrm the validity of the results.

1.6 Dissertation structure

Chapter 2 introduces some of the recent approaches proposed for ecient Medium Access Control (MAC) in Vehicular Ad-hoc NETworks (VANETs) and provides an in depth analysis of work in this area. The existing IEEE WAVE standards, IEEE 802.11p and IEEE 1609.1-4 are explored, and the problems associated with this approach are high-lighted. An analysis of some of the recently proposed alternative MAC approaches and their strengths and weaknesses are provided. The MAC approaches are categorised as either contention-based, or contention-free, and the contention-free approaches as either group-based or distributed.

Chapter 3 introduces the tools used to simulate vehicular communications in this disser-tation. The conguration and set-up of the simulation tools are presented. The chapter explains how a trac simulator, SUMO, and real-word maps can be used to simulate vehicular movement. A communications simulation platform called OMNeT++, and a wireless simulation framework called MiXiM are presented. The communications model and the trac models used for the simulations are presented. The Chapter also contains an overview of metrics used in the literature for simulation and analysis of MAC layers, and concludes with a description of, and justication for, the metrics used in this dissertation, namely packet delivery ratio, receiver throughput, and end-to-end

(27)

latency.

Chapter 4 introduces three MAC methods. The rst MAC method, called Medium Ac-cess through Memory Bifurcation and Administration (MAMBA), demonstrates how two simple changes in the awareness and memory management of an existing MAC can result in signicant performance improvements. The second MAC method, called Neighbour Yielding and Aware Medium Access (NYAMA), uses three concurrent pro-cesses to manage awareness and awareness memory. An analytical model of NYAMA is developed to predict the eect of various parameters (receiver sensitivity, path loss variations, and vehicle density) on the performance in terms of the chosen metrics. Special focused is placed on the eect of the range and the memory size used for neighbour awareness. Finally, NYAMA is also extended to a neighbour-agnostic MAC to evaluate the eect of removing awareness from the MAC.

Chapter 5 presents the simulation test results of the three proposed MAC methods, using simulation platforms and the performance metrics (receiver throughput, packet delivery ratio, and end-to-end latency) described in Chapter 3. The simulation results are presented with the analytical results from Chapter 4.

Chapter 6 concludes the work by validating the hypotheses from Chapter 1 using the results from Chapter 5. The main ndings and contributions of the work in this dis-sertation are provided.

(28)

Chapter 2 Literature Survey: MAC Approaches

2.1 Introduction

This chapter provides a detailed survey of the literature on Medium Access Control (MAC) for vehicular communications. We rstly provide an overview of the standards proposed by the IEEE, as well as a brief history of how these standards evolved. We then critic-ally review the MAC support proposed by the IEEE, followed by alternative approaches recently proposed in the literature. The work in this chapter has been published as [9].

2.2 Dedicated Short Range Communication (DSRC)

and Wireless Access for the Vehicular

Environment (WAVE)

Dedicated Short Range Communication (DSRC) [31] was initially used in Europe to describe the protocol used for only Electronic Toll Collection (ETC). These systems are now in use worldwide, in countries that include China, Australia, and South Africa. Current DSRC systems, mostly used for tolling, comply with dierent and incompatible standards in Japan, Europe, and the US. Current DSRC ETC systems are based on the European Committee for Standardisation (CEN) standards EN 12253, EN 12795, EN 12834 and EN 13372. EN 12253 uses the frequency ranges 915 MHz and 5.795 5.815 GHz. In the US however, DSRC and Wireless Access for Vehicle Environments (WAVE) [32] are interchangeably used to broadly describe vehicular network technology, based on the on the IEEE 802.11 standard. In Europe this is referred to as Intelligent Transportation Systems (ITS) [33].

(29)

In the United States, the DSRC spectrum is regulated by the US Federal Communic-ations Commission (FCC) [34]. The European Telecommunications Standards Institute (ETSI) and the EU are responsible for regulating usage of this spectrum and published a standard, EN 302 571 2008, which allocates the same frequency range as DSRC in the US [25]. The American Society for Testing and Materials (ASTM) developed a single standard for the PHYsical layer (PHY) and the Medium Access Control (MAC) called ASTM E 2213 [35] with the purpose of dening the over-the-air radio frequency protocol for a DSRC system. Further work on this ASTM E 2213 standard has recently been un-dertaken by task group p of the IEEE 802.11 working group. The IEEE has subsequently released standards P1609.1 through P1609.4 and standard IEEE 802.11p, which is a deriv-ative of IEEE 802.11a, for initial assessment in VANETs, which they call Wireless Access for Vehicle Environments (WAVE) [19]. These standards are summarised in Table 2.1 [32, 36, 1].

The next generation of DSRC, as dened by the IEEE WAVE standards and the ETSI standards, uses the 5.9 GHz band. The US DSRC channel assignment, as per the U.S. Code of Federal Regulations (CFR) for telecommunications title 47 parts 90 [37] and 95 [38], is listed in Table 2.2 [39]. WAVE allocates seven channels for vehicular communications. One channel is designated as a Control CHannel (CCH) for emergency and coordination, four channels as Service CHannels (SCHs), and a channel each for critical messages and high power public safety messages.

Figure 2.1 summarises the IEEE standards used for the vehicular communications environment, and Figure 2.2 shows how the IEEE standards and layers relate to the well-known ISO layers. [19, 1, 23,40].

2.2.1 Europe

The EU spectrum allocation authority has allocated a frequency range of 30 MHZ, 5.875 - 5.905 GHz for ITS safety use in [41], with the intention to increase the range to 70 MHz. Additionally, in Europe, the International Organisation for Standardisation (ISO) set up Technical Committee (TC) 204 that is made up of a number of working groups. The

Table 2.1: IEEE standards for VANETs. IEEE Purpose and function

1609.1 WAVE resource manager

1609.2 Security, secure message formatting, processing and message exchange 1609.3 Routing and transport (networking) services (alternative to IPv6).

Provides management information base for the protocol stack

1609.4 Multiple-channel operation in the DSRC standard, supplementing the IEEE 802.11p

802.11p Specication for the physical and MAC layer to enable operation in the WAVE (highly mobile nodes), based on 802.11a

(30)

WAVE Resource Manager - IEEE 1609.1

MAC & PHY - IEEE 802.11p Multi Channel Operation - IEEE 1609.4

TCP/UDP & IP WSMP & WME - IEEE 1609.3 Security

IEEE 1609.2

IP Internet Protocol PHY PHYsical layer UDP User Datagram Protocol WME WAVE Management Entiry

MAC Medium Access Control TCP Transmission Control Protocol WAVE Wireless Access Vehicular Environment WSMP Wave Short Message Protocol

Figure 2.1: Overview of the IEEE VANET standards. IEEE 1604.1-4 and IEEE 802.11p de-scribes the communications stack used for Wireless Access in the Vehicular Environment (WAVE) [32] 2.4

Table 2.2: DSRC Channel allocation by the FCC. Channel number Frequency (MHz) Description

172 5855-5865 Critical, safety of life

174 5865-5875 Service channel

178 5885-5895 Control Channel

184 5915-5925 High power, public safety

ISO TC has developed a framework for ITS called Communications Access for Land Mo-biles (CALM) encapsulating VANET standards from IEEE, ETSI, CEN and the Society of Automotive Engineers (SAE). CALM includes various types of wireless technologies including WiMAX, Wi-Fi, EDGE, GPRS, 2G, and 3G. The CALM equivalent in the 5.9 GHz range is called CALM M5 and occupies the same frequency range as DSRC in the US.

2.2.2 Japan

In Japan, the Association of Radio Industries and Businesses (ARIB) standard T75 is used for ETC. This standard prescribes the frequency range 5.77 - 5.85 GHz, with the uplink and downlink channels separated by 40 MHz.

The frequency ranges for DSRC for dierent regions of the world alongside the ITU-R Industrial Scientic and Medical (ISM) radio band are summarised in Figure 2.3

(31)

OSI

WAVE

Layer 1 Physical Layer Layer 2

Data Link Layer Layer 3

Network Layer Layer 4

Transport Layer WME 1609.3 PLME 802.11p MLME 802.11p MLME Extension 1609.4 Resource Manager 1609.1 Safety Applications Non-safety Applications Traditional DSRC PHY 802.11p MAC 802.11p Multi-Channel Operation (Upper MAC) 1609.4 LLC IP TCP UDP WSMP 1609.3 _Security 1609.2

DSRC Dedicated Short Range Communication

LLC Logical Link Control MLME MAC Layer Management Entity PLME Physical Layer Management Entity

WME WAVE Management Entity WSMP Wave Short Message Protocol

Data Management

Figure 2.2: IEEE VANET standards.

5.72 5.77 5.82 5.87 5.92

Europe ISO CALM (CALM M5) Europe ETSI ITS (Decision 2008) EU Safety (EN 302 571 2008) North America DSRC (IEEE 802.11p) Japan ETC (ARIB T75) Europe DSRC ETC CEN (EN 12253) ITU-R (ISM band)

GHz

Figure 2.3: VANET and conventional DSRC frequency allocations for dierent parts of the world.

2.3 IEEE WAVE MAC Standards IEEE

802.11p/IEEE 1609.4 Standards

Network synchronisation in IEEE 802.11p networks is achieved by dividing channel time into repeating intervals of 100 ms. Every interval has a Control CHannel (CCH) and a Service CHannel (SCH) allocation, both with a guard band [32]. The channel synchron-isation scheme specied in the IEEE 1609.4 standard is illustrated in Figure 2.4. During the CCHI Interval (CCHI) all nodes tune to the CCH to listen for emergency messages and advertisements of services on other channels. If a node is interested in a service or information on an SCH channel, it tunes to that channel during the Service CHannel

(32)

Interval (SCHI).

The contention-based scheme employed in IEEE 802.11 networks, including IEEE 802.11p, is CSMA/CA, and the basic MAC technique used by 802.11 networks is the Distributed Coordination Function (DCF). When a node has a packet to transmit, it senses the medium until it detects an idle state (i.e. no transmissions) for a period called the DCF Inter-Frame Space (DIFS). After sensing such a period of inactivity on the medium, the node selects a counter value for an additional waiting period called the Contention Window (CW). The counter value is selected randomly, but within bounds of a set minimum and maximum called CWMIN and CWMAX. While the medium is idle, the CW counter counts down until it either reaches zero or senses activity on the medium. Every transmit attempt expects an ACKnowledge (ACK) to conrm successful transmission. The CW is doubled in the event of unsuccessful transmission, up to the maximum value.

To avoid the hidden or exposed terminal problems, a Request To Send, Clear To Send (RTS/CTS) handshaking method is employed along with a Network Allocation Vector (NAV) [42]. Each node maintains a NAV, which is extracted from the RTS, CTS, Data and ACK packets. NAV acts as a virtual carrier sense by predicting when the medium will be busy and not transmitting then (similar to normal carrier sense that waits until it senses a free medium before attempting to transmit).

Coordination in the broadcasting mode of IEEE 802.11p is simpler: If a unit wants to broadcast, it waits for a period called the Arbitration Inter-Frame Space (AIFS). If no activity is sensed during the AIFS, the the node can transmit. However, if activity is sensed during the AIFS, the node that wants to transmit backs o for a period called the Contention Window (CW), after which the process starts again with the AIFS. The CW is based on a counter, which increases if the CW is entered into again before a successful

Service SCHI 50 ms CCHI 50 ms Sync Interval 100 ms Control CSMA/CA CSMA/CA CSMA/CA CSMA/CA CSMA/CA CSMA/CA CSMA/CA Ch 1 Ch 2 - 7 Broadcast Broadcast UTC second Service Control Sync Interval 100 ms

CCHI Control Channel Interval CSMA Carrier Sense Multiple Access CA Collision Avoidance

ms milliseconds

SCHI Service Channel Interval TDMA Time Division Multiple Access UTC Universal Time, Coordinated

SCHI 50 ms CCHI

50 ms

Control Channel Interval Carrier Sense Multiple Access Collision Avoidance

milliseconds

Service Channel Interval Time Division Multiple Access Universal Time Coordinated

(33)

transmission attempt.

In contrast to the conventional IEEE 802.11 network, IEEE 802.11p does not contain authentication and association in the MAC and PHY layers. This is because the normal modes of authentication and association would not meet the stringent timing requirements set by the VANET environment (for example for the scenario where two vehicles move at normal speed in opposite directions). Moreover, the notion of the 802.11 Basic Service Set (BSS) is replaced in 802.11p with a WAVE-BSS (WBSS). In a conventional 802.11 network, a BSS is the collection of nodes that are connected and are able to communicate. Two types of BSSs could exist in a conventional 802.11 network, namely an Infrastructure BSS, which includes an Access Point (AP), or an independent BSS, which is formed without an AP on an ad-hoc basis. For WAVE, however, any node is allowed to transmit in a WBSS before any authentication or association, as long as the node has received a WBSS announcement from a WBSS provider [19, 43].

IEEE 802.11p uses Enhanced Distributed Channel Access (EDCA) functionality, de-rived from the IEEE 802.11e standard [1]. EDCA is based on Carrier Sense Multiple Access with Collision Avoidance (CSMA/CA) and improves Quality of Service (QoS) by introducing dierent message priorities. Prioritisation is achieved by varying the Conten-tion Windows (CWs) and the Inter-Frame Spaces (IFS), which changes the probability of successful medium access. The EDCA resides in the MLME Extension and allows messages with a higher priority (there are 4 categories) to have a better chance of being transmitted than messages with a lower priority. To achieve this, the Contention Win-dow (CW) and the Arbitration Inter-Frame Space are shortened. EDCA also provides contention-free periods, called Transmit OPportunities (TXOP).

IEEE 1609.4 species the multiple channel operation for the MAC and PHY of an IEEE 802.11p WAVE system using the Control Channel (CCH) and Service CHannels (SCHs). It also provides for prioritisation, routing, and coordination. The WAVE service advertisements and channel coordination are to be performed on the control channel (Channel 178 as designated by the FCC) [39, 25]. QoS cannot be guaranteed for safety critical messages and other real-time transmissions [27].

2.4 Contention-free approaches

This section presents a critical review of the contention-free MAC approaches recently proposed for use in VANET.

2.4.1 MAC scheme for fair access.

In [20], a contention-based MAC for V2I is developed by Karamad et al., with the purpose of increasing access fairness. Although this MAC is developed for RSU-based communic-ation, it is included it in this work to highlight the alternative interpretation of fairness and the resulting MAC approach. Since vehicles do not travel at the same speed, yet have

(34)

an approximately equal communication range, all vehicles do not have the same time to access a RoadSide Unit (RSU). Furthermore, faster nodes are less likely to be able to com-municate with a given RSU than slower nodes in the same travel path. The approach is based on IEEE 802.11, with the DCF adjusted for node speed. In the proposed MAC ap-proach, the contention window of each vehicle is increased for higher vehicle speeds, which enables fairer access of the shared medium. The approach has a number of shortcomings and is not suited for V2V communication: A high level of coordination, awareness, and overhead are required to be able to actively adjust the DCF for each node relative to other nodes. Moreover, the speed of other nodes is only known once within range. In the V2V scenario, contact time is already very short. It is dicult to justify the cost of data transfer time given the small benet or increased fairness as dened in [20].

2.4.2 Self-organising TDMA (SoTDMA)

One of the popular proposed MAC methods is a Self-organising TDMA-based (SoTDMA) MAC method, evaluated in the vehicular communication environment by Bilstrup et al. in [24] for safety-critical message delivery. The method is presently employed in aviation and naval surveillance as part of the Automatic Identication System (AIS) and VHF Data Link Mode 4 respectively [44, 45]. A repeated sequence of a xed number of slots, called a frame, is used, and every node synchronises its slots to the Global Positioning System (GPS) time. The frames are not synchronised. Every node selects a range of slots to choose from its frame and at random intervals chooses a dierent and unoccupied slot from this range. The frame structure is illustrated in Figure 2.5.

How often a node transmits is dependent on its speed: the faster it goes, the more it transmits. During a slot, the node that selected the slot transmits the positional information of the transmitting node and an optional short emergency message. SoTDMA AIS was developed for very low data rates to announce only positional information (a few bytes), and not for safety message payloads typically expected to contain supplemental information ([30] suggests 10 kilobits) to avoid accidents and control trac. Similarly, the approach is designed for slow moving maritime vessels and might not be suited for vehicles with high mobility. However, the work in [24] demonstrates that some optimization to the approach results in acceptable results in the vehicular environment, and for vehicular

Frame Start

Range Range Range

Repeat Interval Repeat Interval

Chosen

Slot Chosen Slot

Chosen Slot

Frame

Figure 2.5: Message structure in Self-organising TDMA. Each node selects a range of slots in the frame, from which it randomly selects free slots to transmit in.

(35)

applications. If the slots become saturated, nodes that are furthest apart start to share slots. If these slots are not suciently separated in space, collisions will occur, which could lead to unbounded delays.

2.4.3 Vehicular Self-Organising MAC (VeSOMAC)

A self-organising contention-free TDMA MAC, called VeSOMAC was developed and eval-uated by Yu et al. [27]. The goal of the research was to develop a contention-free MAC method with distributed control, which delivers increased data transfer between vehicles that drive in platoons (groups) in highway scenarios. VeSOMAC can operate with all nodes synchronised with GPS time, or with nodes self-adjusting asynchronously. We will focus on the asynchronous operation, since that is what [27] describes in detail. The mes-sage structure for three nodes (X, Y, and Z) is illustrated in 2.6. Similar to the SoTDMA in [24], every slot is allocated an equal time duration within a repeating superstructure called a frame. All frames have the same duration, but are not synchronised to that of other nodes. Every node sends a bitmap vector during its allocated transmission oppor-tunity announcing which slots are being used by its one-hop neighbours. The bitmap vector only contains 1's and 0's to indicate the occupancy of slots, and does not the iden-tity of the occupying nodes.

Since frames as seen by each node are not synchronised (for example to GPS time) and

Z Y Z

X Occupied

(by Z) Occupied

(by Z)

The frame according to node X Occupied (by Y) Occupied (by Y) Occupied (by X) Y Occupied (by X)

The frame according to node Y

Occupied (by Z) Occupied (by Z) Z Occupied (by Y) Occupied (by X) Occupied (Y/Z)

The frame according to node Z

X All transmissions relative to node X

1 0

0 1

Bitmap vector transmitted by node X 1

0 1

0 0

Bitmap vector transmitted by node Y

Occupied (by Z)

0 1

Bitmap vector transmitted by node Z 1

0 0

1

Figure 2.6: Vehicular Self-Organising MAC (VeSOMAC) message structure for three nodes. Nodes transmit a bitmap to indicate which slots are used by its neighbours. Transmission is asyn-chronous, which necessitates the use of one bit to represent two slots leading to underutilisation of the medium.

(36)

could drift apart by as much as a full slot, a worst case synchronicity (marginally less than two slots) has to be assumed (Figure 2.6). Each bit in the bitmap vector therefore eect-ively represents two slots to avoid collisions, which leads to underutilisation of slots. Since all nodes send the transmission slots of all their one-hop neighbours, it is claimed that a new node can avoid collisions with all its two-hop neighbours by choosing a slot that is unoccupied according to the combination of all one-hop neighbour time slots advertised. It is not clear from the paper how a node combines the unsynchronised bitmaps from other nodes to form this timing-dependent decision, since the new node has to assume the worst-case synchronicity between all neighbouring nodes, which signicantly reduces the available slots.

Every vehicle continuously attempts to reallocate its own transmission slot to follow im-mediately after (in time) the node travelling in front of it (in space). For this purpose, GPS information is added to the header transmitted in every allocated slot. Slot relo-cation is done by means of a collision resolution mechanism. When a node repeatedly detects that its neighbours do not acknowledge its allocated transmission slot, it assumes a collision is taking place and reallocates its transmission slot.

The approach has some shortcomings: The system is designed for highway scenarios, and the evaluation is accordingly done for highways only. The system performance is expec-ted to degrade signicantly under urban conditions with numerous vehicles travelling in varying directions. The assumption is made that clocks in each vehicle, even though asyn-chronous, are not drifting relative to each other. Drifting could have the eect that node transmissions collide during one interval and not during the next, thereby not reaching the collision detection threshold necessary to reallocate a new slot to avoid collisions. The scenario where two nodes collide and then iteratively and repeatedly jump to the same transmission slot in an attempt to relocate is not addressed. The size of a frame is a design parameter and does not cater for a congested group seceding to form a new group. The frame will therefore have to be designed for full capacity all the time, leading to severe underutilisation if only a few nodes are present. In addition, there is no provision for utilising the multiple channels allocated by the FCC [39]. The use of a bitmap severely limits the performance of this approach because it obscures the identity of the occupying node and also hides its range. Removing outdated information and making intelligent decisions on the bitmap information is therefore dicult. In addition, when evaluated against IEEE 802.11p, the approach gives higher round-trip delays 2.6.

2.4.4 Multi-Channel Token Ring Protocol (MCTRP)

Bi et al. [28] proposed and evaluated a contention-free Multi-Channel Token Ring Protocol (MCTRP) for VANET. The research goal was to develop a contention-free MAC method that autonomously organise nodes into token passing rings to achieve low latency for safety messages and increased network throughput for non-safety applications. MCTRP works by grouping nodes with similar velocities into rings, each with a founder-leader node. A

(37)

token passing TDMA scheme is used to control access to the medium for intra-ring data transmission, and CSMA/CA is used to control access to the medium for inter-ring data, emergency, and ring administration transmissions. Every node is equipped with two ra-dios. The MCTRP message structure is illustrated in Figure 2.7. One is permanently tuned to WAVE channel 178, as allocated by the FCC [39], for inter-ring data communica-tion, inter-ring safety message transmission and ring set-up. The second radio is tuned to one of the six remaining WAVE channels that every ring uses for intra-ring transmissions of safety, coordination and data messages. MCTRP employs GPS synchronisation and partitions its equal and repeating time segments T into safety periods Ts, coordination

periods Tc, and data exchange periods T d, as illustrated in Figure 2.7. This timing

se-quence is shared between Radio 1 and 2. MCTRP suers from the following weaknesses: the ring topology along with the ring founder-leader nodes make the system heavily reli-ant on these nodes. If a ring leader node departs, or goes out of range, the ring collapses and association has to be re-initiated. This makes the MAC more suited to scenarios where vehicles tend to platoon, rather than more erratic and unpredictable mobility pat-terns. The rules for ring organisation and channel communication are computationally expensive, especially when faced with heavy trac load and high node mobility. The ring topology is also fairly static since ring size is xed. Given the ring topology, xed ring sizes and xed joining speed thresholds, it is likely that many nodes may not be able to join the rings. The system relies on connectivity between all nodes in the ring since the token needs to be passed between nodes and the founder-leader needs to be aware of all interactions. This scenario is idealistic and not likely because of the high mobility of vehicles, varying separation between vehicles, and harsh communication conditions. MC-TRP uses CSMA/CA for safety messages and token passing TDMA for data transfers. This leads to a scenario where safety messages could face unbounded delays under heavy loads. The system depends on GPS for external timing as well as positional information. MCTRP is dependent on two radios per vehicle and full transmission slots are used even though a node may not need to transmit.

Inter-ring communications and ring founding - Radio 1 on CH 178

Intra-ring - Radio 2 on one of CH {172, 174, 176, 180, 182} as per RFN instruction CSMA/CA

CSMA/CA

Token passing ring topology, coordinated by RFN during Tc

CSMA/CA with RTS/CTS.

CSMA/CA CSMA/CA with some NAV

Repeated time frame (T) synchronized to GPS

Safety (Ts) Coordination (Tc) Data Exchange (Td)

Figure 2.7: Multi-Channel Token Ring Protocol (MCTRP) Message Structure. A token passing TDMA scheme is used to control access to the medium for intra-ring data transmission, and CSMA/CA is used to control access to the medium for inter-ring data, emergency, and ring administration transmissions.